Models, solutions and enabling technologies in humanitarian ...

Viewer
Transcript

European Journal of Operational Research 244 (2015) 55–65

Contents lists available at ScienceDirect

European Journal of Operational Research journal homepage: www.elsevier.com/locate/ejor

Models, solutions and enabling technologies in humanitarian logistics Linet Özdamar a,∗, Mustafa Alp Ertem b,1 a b

˙ Industrial and Systems Engineering Department, Yeditepe University, 26 A˘gustos Yerlesimi ¸ Kayısda˘ ¸ gı Cad., 34755 Istanbul, Turkey Department of Industrial Engineering, Çankaya University, Eskisehir ¸ Yolu 29.km, 06810 Ankara, Turkey

a r t i c l e

i n f o

Article history: Received 13 February 2014 Accepted 18 November 2014 Available online 26 November 2014 Keywords: Humanitarian logistics Modeling approaches Information systems

a b s t r a c t We present a survey that focuses on the response and recovery planning phases of the disaster lifecycle. Related mathematical models developed in this area of research are classiﬁed in terms of vehicle/network representation structures and their functionality. The relationships between these characteristics and model size are discussed. The review provides details on goals, constraints, and structures of available mathematical models as well as solution methods. In this review, information systems applications in humanitarian logistics are also surveyed, since humanitarian logistics models and their solutions need to be integrated with information technology to enable their use in practice. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Since the 1950s, the number and magnitude of disasters have grown exponentially, the number of affected people has grown in proportion (about 300 million persons per annum on the average since the 1990s) and the annual damage costs have risen to about 0.17 percent of the world GDP (Guha-Sapir, Hoyois, & Below, 2014). The International Disaster Database indicates that Asia and the Americas are the most affected continents by disasters (mostly by ﬂoods, quakes and storms) with Europe standing as the third most affected continent. Recently, researchers have conducted various surveys on disaster management. Altay and Green (2006) emphasize lifecycle phases in disaster operations management, and later on, Apte (2009) discusses details of solution methods and limitations of proposed models designed for each phase. Natarajarathinam, Capar, and Narayanan (2009) provide a review for supply chains in times of crisis in both OR/MS and supply chain management journals. Caunhye, Nie, and Pokharel (2012) review optimization models for pre-disaster facility location, stock prepositioning, relief distribution, and casualty transportation. de la Torre, Dolinskaya, and Smilowitz (2012) focus on relief routing models with an emphasis on egalitarian and utilitarian objectives in disaster response. Galindo and Batta (2013) compare and contrast the articles in Altay and Green (2006) with recent developments in humanitarian logistics to highlight whether OR/MS researchers have addressed the aforementioned literature gaps.

∗

Corresponding author. Tel.: +90 216 578 04 50, +90 216 5780 000; fax: +90 216 578 04 00. E-mail addresses: [email protected] (L. Özdamar), [email protected] (M. A. Ertem). 1 Tel.: +90 312 233 1367; fax: +90 312 233 1026. http://dx.doi.org/10.1016/j.ejor.2014.11.030 0377-2217/© 2014 Elsevier B.V. All rights reserved.

Anaya-Arenas, Renaud, and Ruiz (2014) discuss location and transportation models and categorize them according to their objectives, constraints and solution methods. We observe that in these surveys, recovery phase models are not as emphasized as preparedness and response models. Here, we review the logistics models developed for the response and recovery planning phases in terms of their modeling features and formulation structures. Selected articles describe various models that are typical with respect to the latter features. We discuss vehicle representation styles that impact model solvability in capacitated vehicle routing models and categorize relevant models accordingly. We regret that we cannot cite all available papers pertaining to each category due to space limitations. We also investigate the technological advances that facilitate the execution of proposed models and solutions on various types of information systems. The latter review leads to the observation that standard software that integrates different planning phases is needed in order to achieve unobstructed international humanitarian cooperation. The goal would be to enable global access to such software by all countries sharing emergency resources. We end the survey with concluding remarks that point out the areas that bear more improvement. We note that this survey overlaps partially with the one presented in a keynote speech (Özdamar, EURO-INFORMS Conference, July 2nd, 2013, Rome). 2. Concentration areas in humanitarian logistics research Humanitarian logistics research is dedicated to the following three planning stages in the disaster lifecycle: (pre-disaster) preparedness phase, (post-disaster) response and recovery phases. In the pre-disaster phase, preparedness plans and risk-prevention actions, such as infrastructure and building re-enforcement, reduce

56

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

Table 1 Models for relief delivery and casualty transport. Citation

Model type

Objective(s)

Constraints

Haghani and Oh (1996) Özdamar et al. (2004)

DNF DNF

OC CUD

CV, D, MC, MD, SLD, TW CV, D, LS, MC, MD, MM, SLD

Yi and Özdamar (2007)/Özdamar and Yi DNF (2008)/Yi and Kumar (2007) Zhan and Liu (2011) DNF

WUD (CUD + CUC)

Afshar and Haghani (2012) Najaﬁ et al. (2013)

WUD (CUD + CUC) (CUD + CUC) + NV

ETT, EUD

Özdamar (2011) Özdamar and Demir (2012) Zhang et al. (2012)

DNF DNF, robust optimization SNF SNF UNF

Barbaroso˘glu and Arda (2004)

UNF

TT TD TT (primary and secondary disasters) TC

Tzeng et al. (2007)

UNF

EQD, TC, TT

Gu (2011) Balçık et al. (2008) Lin et al. (2011) Hsueh et al. (2008) Barbaroso˘glu et al. (2002) de Angelis et al. (2007) Shen et al. (2009)

UNF (fuzzy) RE, dynamic RE, dynamic CVR CVR CVR 1st stage: stochastic CVR 2nd stage: LP CVR: goal prog. CVR CVR, covering TSP –

Vitoriano et al. (2009, 2011) Berkoune et al. (2012) Nolz et al. (2011) Sheu (2010)

Solution methods

Lagrangean relaxation Lagrangean relaxation, modiﬁed shortest path CV, DP, LS, MC, MD, MF, MPS, MPQ, SLD Exact/tour construction/Ant colony algorithm CV, D, DU, LS, RU, SU Two-stage stochastic program with recourse CV, D, LS, MC, MD, PL, SLD Exact CU, CV, DP, DU, LS, MC, MD, MF, MM, Exact SLD, SU CV, DP, LS, MD, MC, MF, PAY, RE, SLD Exact CV, DP, LS, MC, MD, MF, SLD Hierarchical planning, clustering D, LS, MC, MD, UV LP based heuristic D, DU, MC, MD, MM, RU, SU, UV

Two-stage stochastic program with recourse Exact

D, LS, MC, MD, SLD, UV, 2 stage supply chain UD LS, TTU, TW, UV TC, minimax CUD(EQD) CV, D, LS, MC, MD, SD, TW EQD, TT, UD CV, D, MC, SD, SLD, TW, WTW TT, LA CV, DP, TTU, TW, dynamic scheduling OC, RT CV, DP, MC, PAY, RE, SD, SLD SD CV, D, MD, PL, TW UD, AT CV, D, DU, RTD, SD, SLD, TTU

Exact Exact Decomposition, GA Route construction heuristic Hierarchical planning Exact Exact, Tabu Search

EQD, OC, RR, RS TT AP, MCD, TT, UN SD

Exact Set enumeration heuristic, GA Memetic algorithm Fuzzy clustering heuristic

BUD, CV, DP, MD, STE CV, D, MC, MD, WTW CV, PC, SD, STE D, DU, UV

Objectives – AP: number of alternative paths, AT: vehicle arrival time, CUC: cumulative unmet demand over time, CUD: cumulative unserved casualties, EQD: equity of satisﬁed demand, ETT: Expected Travel Time, EUD: expected unmet demand, FT: ﬂow time, LA: late arrivals, MCD: maximal demand covering, NV: number of vehicles in transit, OC: operation cost, RR: road reliability, RS: road security, RT: response time, SD: satisﬁed demand, TC: transport cost, TD: travel distance, TT: travel time, UD: unmet demand, UN: unreachability of nodes, WUD: weighted unmet demand; constraints – BUD: budget for vehicles, CU: casualty uncertainty, CV: capacitated vehicles, D: delivery only, DP: delivery and pickup, DU: demand uncertainty, LS: limited supplies, MC: multicommodity, MD: multi-depot, MF: medical facilities, MM: multi-mode transport, MPS: medical personnel sharing, MSQ: medical service rates, PAY: helicopter payload, PC: population covering, PL: parking limitation, RE: refueling, RTD: response time deadline, RU: road uncertainty, SD: single depot, STE: sub-tour elimination, SU: supply uncertainty, TTU: travel time uncertainty, TW: delivery time windows, UV: uncapacitated vehicles, WTW: working time windows, SLD: split deliveries (pickups); model types – DNF: dynamic network ﬂow, TS: Time-Space Network, UNF: uncapacitated network ﬂow, SNF: static network ﬂow, CVR: classical vehicle routing, RE: route enumeration.

damage in disaster prone areas. Furthermore, inventory and equipment pre-positioning enable fast relief distribution to survivors from pre-stocked commodities (Balçık & Beamon, 2008; Duran, Gutierrez, & Keskinocak 2011; Ichoua, 2010; Mete & Zabinsky, 2010; Ukkusuri & Yushimito, 2008). On the other hand, pre-positioning of shelters leads to coordinated evacuation of the population to nearby shelters (Widener & Horner, 2011). Pre-disaster models are usually stochastic due to the uncertainty of the disaster impact level. Two-stage stochastic programming (Mete & Zabinsky, 2010) and stochastic scenario analysis approaches (Balçık & Beamon, 2008; Chang, Tseng, & Chen, 2007) are frequently utilized to solve these models. Post-disaster phases involve mass evacuation to shelters (can also be executed as a pre-event activity if notice of disaster is given), relief delivery, casualty transportation, debris collection, road clearing and regional re-development. These activities enhance post-disaster survival rates and economic growth. We review response and recovery logistics models in the following sections.

2.1. The response phase Response phase logistics models fall under two major categories: a) relief delivery/casualty transport models, b) mass evacuation models. Table 1 provides a list of models related to relief delivery/casualty transport with regard to their objectives, model particulars and structure, decisions considered, and solution methods. Table 2 describes mass evacuation models in a similar fashion. In the following sections, we discuss several vehicle representation styles involving capacitated

vehicle transport, and then, we classify the literature according to the latter feature and other characteristics. 2.1.1. Vehicle representation styles in models with capacitated vehicles Relief delivery, casualty transport and public transit mass evacuation models are involved with vehicle routing. In relief delivery, models consider the transport of supplies to demand points by capacitated vehicles. Vehicles start their itineraries at warehouses and serve several demand points on their routes. A trip does not necessarily end at the point of origin. In casualty transport, vehicles visit several locations where casualties are waiting, pick them up and transfer them to one or more medical facilities. In mass evacuation, where threatened populations have to relocate to safe zones, evacuation may take place in personally owned cars or public transit transport. The ﬁrst type of evacuation involves traﬃc ﬂow management rather than individual vehicle routing. The second type of evacuation involves picking up waiting evacuees at bus stops and transferring them to shelters. In this case, bus capacity and vehicle routing issues arise. Vehicle routing in humanitarian logistics context often involves split deliveries and pickups due to the magnitude of demands. Models that consider capacitated vehicles have restrictions related to vehicle capacity, vehicle-load conformity (such as ambulances for casualties, lorries for food), road reliability, security and accessibility, supply/vehicle availability, depot/shelter/pickup location capacity, delivery time windows, re-fueling and parking limitations, split or non-split deliveries/pickups, response time deadlines, budget limits on vehicle ﬂeet size, and vehicle working time windows. The goals stated are usually cost/travel time-, demand statisfaction-, response

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

57

Table 2 Models for mass evacuation. Citation

Objectives

Model type

Model features

Solution method

Chiu and Zheng (2007)

TTT

CTM(LP), CE

Exact

Liu et al. (2006) Miller-Hooks and Patterson (2004)

Bi-level: MF, TTT NCT

Lim et al. (2012) Cova and Johnson (2003)

TNE (weighted sum of ﬂows) TTD

CTM (LP), CE Time dependent quickest ﬂow (LP), CE Capacitated MCNF (MIP), CE MCNF(MIP), CE

Multi priority group evacuation, managing traﬃc ﬂow, destination selection, departure schedules Traﬃc ﬂow and routing TS network, time variant travel times and link capacities TS network, paths, schedules and ﬂows

Naghawi and Wolshon (2012) Cui et al. (2014)

TNE TCF

Kongsomsaksakul et al. (2005)

TET, TTT

Hobeika and Kim (1998)

TET, NC

Campos et al. (2012)

MF

Chen and Chou (2009) Perkins, Dabipi, and Han (2001) Mastrogiannidou et al. (2009)

NCT, DT, TTT, ATS TTT TTD

Multi modal MCNF, CE Nonconvex nonlinear MCNF, CE MINLP: bi-level location–allocation model, CE Traﬃc simulation model, CE MCNF (LP): maximum ﬂow, CE Simulation, PT Simulation, PT Simulation, PT

Bish (2011) Murray-Tuite and Mahmassani (2003)

TET + MET TTT, WT

CVR, RE, PT CVR, PT

Liu and Yu (2014)

TNE, TTD

CVR, PT

Song et al. (2009)

TTD

Stochastic CVR, PT

He et al. (2009)

TET

Stochastic LR (MIP), PT

Sayyady (2007), Sayyady and Eksioglu (2010) Margulis et al. (2006) Na et al. (2012)

TC, TET, WT

DNF-TS (MIP), PT

TNE MET, TC (secondary evacuation)

RE, PT RE, PT

Lane-based routing, intersection delay management Evacuation routes, ﬂows Time variant travel times, contraﬂows (rescue ﬂows), lane reversal Pre-ﬂood evacuation routes and shelter selection

Optimization and simulation Exact Dijkstra for paths; max ﬂow for ﬂows (greedy) Successive shortest path Simulation Exact GA

Routes and schedules

Simulation

Non-intersecting paths to shelters, link capacities

Iterative heuristic

Bus assignment to routes, single trip, contraﬂows Pre-set routes, bus departure times, single trip Bus assignment to pickup stops based on shortest time, multiple trips Split delivery, multi depot, multiple trip Vehicle routes from their point of start in the network (anywhere) to pickup to shelter Bus stop capacity, evacuee to stop allocation, single trip Demand uncertainty, evac. deadline, shelter selection Shelter selection, bus ﬂeet size, routes, demand uncertainty Pedestrian routes, bus routes, single trip

Simulation Simulation Cluster ﬁrst, route second

Bus assignment to preset routes, multiple trips Route assignment (with secondary evacuation), capacitated buses and hospitals

Constructive heuristic Exact Exact Simulation based GA GA, hill climbing, ANN Tabu Search Exact Exact

Objectives – TTT: Total Travel Time, TTD: Total Travel Distance, TET: Total Evacuation Time, DT: Delay Time, NCT: Network Clearance Time, MET: Maximum Evacuation Time, WT: Waiting Time of evacuees, TCF: Costs of Flows, TC: Travel Cost, ATS: Average Traﬃc Speed, TNE: Total Number of Evacuees, NC: Number of Casualties, MF: Maximum Flow; model types – LP: Linear Program, MIP: Mixed Integer Program, BP: Binary Program, MINLP: Mixed Integer Nonlinear Program, CTM: Cell Transmission Model, MCNF: Minimum Cost Network Flow, CE: Car evacuation, PT: Public Transport, LR: Location-Routing, CVR, RE, DNF-TS as deﬁned in Table 1.

time-, and road-risk related. Among these, egalitarian goals (satisfying all beneﬁciaries equally), and utilitarian goals (minimizing sums rather than mini-max objectives) should be differentiated (Bish, 2011; Campbell, Vandenbussche, & Hermann, 2008; de la Torre et al., 2012; Huang, Smilowitz, & Balçık et al., 2012). Below we discuss the styles in which capacitated vehicles are represented in models. They affect the size of the models solved, and therefore, have an impact on solution times. A. DNF approach. In the Dynamic Network Flow (DNF) approach vehicles are represented as integer valued ﬂows that are linked with commodity ﬂows in a multi-period (dynamic) network ﬂow (NF) model. Vehicle ﬂow variables have link (from–to nodes) indices and a time index that indicate the time of vehicle traversal over a link. i. DNF with time–space (DNF-TS) networks. In DNF-TS models where a time–space (TS) network is used by creating a copy of the network for each time interval in the planning horizon. This structure leads to (T2 N2 ) vehicle related integer variables where T is the length of planning horizon and N is the number of nodes in the network (Haghani & Oh, 1996). ii. DNF excluding time–space networks. In DNF models where travel times are embedded in equations as time lags (Özdamar, Ekinci, & Küçükyazici, 2004), the number of vehicle related integer variables reduce to (TN2 ). In all types of DNF models, once the mixed integer DNF is solved, a post-processing stage has to be activated. At this stage, vehicle routes are constructed

from vehicle and commodity ﬂow information. Yi and Özdamar (2007) accomplish the latter by solving a Linear Program (LP). B. SNF approach. The Static Network Flow (SNF) is a single period NF model applied on a rolling horizon basis. All other features of the SNF are similar to those of the DNF model. The SNF model has (N2 ) integer variables; therefore, it is more eﬃcient. Özdamar and Pedamallu (2011) compare both DNF and SNF models to show that the sizes of SNF instances that can be solved optimally are twice as those of the DNF instances. C. RE approach. The Route Enumeration (RE) approach has a preprocessing stage where all feasible routes between all pairs of supply and demand nodes are constructed. This set of R routes is then fed to a Mixed Integer Program (MIP) as an input vector. Vehicles are then optimally assigned to these routes to satisfy demand. Here, vehicles are represented by binary variables with vehicle, route and time indices. The dynamic RE model has (vRT) binary variables, where v is the number of vehicles (Balçık, Beamon, & Smilowitz, 2008). A setback in this approach is the cardinality of R that can be as large as (2N−2 M (N − M)) where M is the number of depots in a network with N nodes. In large scale relief networks, the construction of R routes becomes computationally intractable and the solution of the MIP requires substantial computation time. In public transit mass evacuation models, the Transportation Department may assign R pre-deﬁned itineraries for all buses. In that case, the cardinality of R is at their discretion. D. CVR approach. In the Classical Vehicle Routing (CVR) approach each vehicle is represented by a binary variable that speciﬁes the link,

58

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

vehicle, and tour indices. This results in (vrN2 ) binary variables, where r is the number of trips allowed for each vehicle. Routes are explicitly constructed by the CVR model and there is no need for pre- or post-processing stages. Yi and Özdamar (2007) compare the CVR model with the SNF model in terms of model size and exact solution time. The researchers observe that the CVR approach inﬂates the model size considerably. As a matter of fact, in emergency logistics, the cardinality of vr can be quite large. 2.1.2. Classiﬁcation of relief delivery and casualty transport literature A. The dynamic/static network ﬂow (DNF/DNF-TS/SNF) models. DNF/DNF-TS models. Models for relief delivery: Cost related goals. Haghani and Oh (1996) consider a DNF-TS model with routing, transfer and inventory carryover links. Vehicle and commodity ﬂows are coupled by vehicle arrival time windows and the goal is to minimize operation costs. The solution involves a Lagrangean Relaxation approach that solves very small networks (10 nodes). Demand satisfaction goals. Özdamar et al. (2004) introduce a multi-modal DNF transport model with limited supplies and dynamic vehicle availability where the goal is to minimize the cumulative unmet demand over T. Two solution approaches, a Lagrangean Relaxation and a modiﬁed shortest path heuristic, are proposed. Afshar and Haghani (2012) model a seven-layer relief delivery supply chain network (the FEMA standard) using a DNF-TS model where features such as storage space, parking space, and inbound/outbound vehicle ﬂow limitations are introduced. The goal is to minimize cumulative unmet demand over T. Integrated models for relief delivery and casualty transport: demand satisfaction goals. Yi and Özdamar (2007) introduce a DNF LocationRouting (LR) model with the goal of minimizing the cumulative prioritized unmet requests. The model selects and locates temporary medical facilities, enables sharing medical staff among facilities, and imposes ﬁnite medical service rates. The researchers solve scenarios with relief networks of up to 60 nodes within 2 minutes of computation time. Yi and Özdamar (2007) solve this model using a two-stage Ant Colony Optimization heuristic where the ants build routes in the ﬁrst stage and commodity ﬂows are optimized in the next by solving the maximum ﬂow problem. Instances with relief networks of up to 80 nodes and 55 vehicles are solved near-optimally. Özdamar and Yi (2008) propose a constructive node-insertion heuristic and solve the same instances faster, but with slightly worse solution quality. SNF models. Integrated models for relief delivery and casualty transport: travel time/distance related goals. Özdamar (2011) proposes a model for coordinating helicopter logistics in medical supply delivery and casualty pickup. The goal is to minimize total ﬂight time plus stopping times. Fuel feasibility and itinerary length limits are taken care of in the post processing stage. The case study involves a relief network with more than 70 helipads scattered over Istanbul (Turkey). Özdamar and Demir (2012) solve large-scale relief delivery and casualty pickup scenarios within a hierarchical planning framework. Recursive clustering is applied to demand nodes in the relief network and the method ensures consistency of supply and hospital service availability at different network aggregation levels. The researchers solve scenarios of up to 1000 nodes near-optimally and verify the efﬁciency of the approach on a Katrina (New Orleans) ﬂooding scenario (Özdamar, Demir, Bakır, & Yılmaz, 2013). B. The Route Enumeration (RE) models. Relief delivery models: cost related goals. Balçık et al. (2008) propose a dynamic relief delivery model with limited supplies and vehicle availability constraints. The goal is to minimize transport, backorder, and lost sales costs where critical items such as tents and blankets are backordered. Regularly consumed items such as food and water become lost sales when demand is not met. In the ﬁrst stage, the researchers identify all possible sets of demand points that can be served in one trip. Each such set represents nodes on a feasible route. The order of nodes on each route is then optimized by solving the TSP. Then, the set of feasible routes

are fed into the model and vehicle assignments are made. Demand and travel time related goals. Lin, Batta, Rogerson, Blatt, and Flanigan (2011) deﬁne three goals in their model: minimize cumulative unmet demand, minimize travel time, and minimize the pairwise difference between demand ﬁll rates at nodes. These are then merged into a single ﬁtness function in a Genetic Algorithm (GA) that enumerates routes. A second solution approach clusters demand nodes ﬁrst and enumerates all routes within each cluster. The GA is found to be slow but its delivery rate is higher than the clustering approach. The clustering approach solves a scenario with three million houses grouped into nine clusters within 2 minutes. C. The Classical Vehicle Routing (CVR) models. Relief delivery models: Demand related goals. de Angelis, Mecoli, Nikoi, and Storchi (2007) schedule cargo planes that carry out food delivery in Angola. In their model, the number of planes and supplies are limited, overnight parking near warehouses is mandatory with limited parking space, split deliveries are not allowed, and take-offs are allowed only in daylight due to security reasons. The goal is to maximize satisﬁed demand. The solution to the problem is obtained in 2 hours by CPLEX. Cost, road security and demand equity goals. Vitoriano, Ortuno, and Tirado (2009 ), and Vitoriano, Ortuno, Tirado, and Montero (2011) propose a multi-objective model for relief delivery. The goals are minimizing transport and operating costs, minimizing the maximum looting probability (based on the number of vehicles traversing arcs), and minimizing the maximum percentage of unmet demand over nodes. The model contains of a transport budget and sub-tour elimination constraints. The Haiti Quake case study is analyzed with three depots, 12 transfer nodes, and nine demand zones to construct the routes of 300 trucks. Berkoune, Renaud, Rekik, and Ruiz (2012) impose a vehicle itinerary length limitation in a model where the goal is to minimize travel distance. A GA is proposed and tested on networks with up to 60 demand nodes and three suppliers. The gap from the best solution is found to be 20 percent. Integrated relief delivery and casualty transport models: Cost and response time related goals. Barbaroso˘glu, Özdamar, and Cevik (2002) propose an integrated CVR model for helicopter pickup and delivery operations and solve very small instances with a two stage hierarchical planning approach. Travel time related goals. Hsueh, Chen, and Chou (2008) use a delivery/pickup CVR model that considers the reverse logistics of relief delivery as well. Travel times are assumed to be time-variant. A node insertion heuristic constructs the initial routes and then improved by node exchange with the goal of minimizing total travel time. A 100 node relief network with ten vehicles is tested. D. Models involving uncertainty. Relief delivery models: Demand and response time related goals. Shen, Dessouky, and Ordóñez (2009) propose a two-stage CVR type model for antidote distribution in large scale bioterrorist attacks, assuming that demand and travel times are uncertain. Their goals are to minimize unmet demand and satisfy service deadline to avoid deaths. The pre-planning stage is modeled by chance constrained programming designed for split delivery vehicle routing and solved by Tabu Search. In the operational stage, the solution of the previous model is ﬁxed for vehicles and a LP determines loads. Tests are conducted on a network of 50 demand nodes. Road risk related goals. Zhan and Liu (2011) consider path availability risk using the NF model and minimize both expected travel time and unmet demand. An infrastructure risk in terms of road network uncertainty is discussed in Nolz, Semet, and Doerner (2011). The problem is to locate and deliver water dispensers so as to cover as much of the population as possible. Road risks are minimized by the following goals: minimize the maximum hazard over all arcs of each path, maximize the number of alternative paths, minimize the unreachability of all alternative paths, and maximize the number of paths whose risk value is below a threshold. The researchers model the problem as a TSP

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

with a single capacitated vehicle and solve it using a multi-objective Memetic Algorithm. Integrated relief delivery and casualty transport models: Demand and congestion related goals. Najaﬁ, Eshghi, and Dullaert (2013) add the goal of minimizing the number of traveling vehicles to the cumulative unmet demand objective in a multi-objective DNF risk model where demand uncertainty exists. Robust Optimization method is used to solve each objective hierarchically. A small case study involving 16 nodes, ﬁve hospitals, and four vehicle types is solved. E. Network ﬂow models without capacitated vehicles (UNF). In Table 1, we also cite UNF relief delivery models that consider cost related (Barbaroso˘glu & Arda, 2004), demand satisfaction related (Gu, 2011), and time, cost and equity based (Tzeng, Cheng, & Huang, 2007 ; Zhang, Li, & Liu, 2012) objectives. Demand, supply and road uncertainties are considered by Barbaroso˘glu and Arda (2004) who propose a twostage stochastic program with recourse to minimize costs. Tzeng et al. (2007) propose a deterministic two stage supply chain with split deliveries and limited supply availability. Gu (2011) considers travel time uncertainty and delivery time windows. 2.1.3. Mass evacuation models Mass evacuation models concentrate on Car Evacuation (CE) and traﬃc ﬂow management, and also, on evacuation by public transit (PT). CE models are frequently represented by link capacitated Minimum Cost Network Flow (MCNF) models with vehicle ﬂow per hour limitations over links. Traﬃc simulation tools are also used to observe the impacts of lane reversals, intersection delay management, and traﬃc direction reversals on traﬃc ﬂows. Some models consider contraﬂows of rescuers. PT models optimize bus routes consisting of a depot, several pickup stations and a shelter. These models consist of pickup only vehicle routing problem. In some models bus departure times are also considered to relieve traﬃc congestion. Shelter selection may also become a part of the problem leading to a Location-Routing (LR) model. In no-notice evacuation, a bus conducts a single trip, however, in short notice evacuation, it can carry out multiple trips. The objectives in PT models can be time related, evacuee number related, cost and ﬂow related. Vehicle representation is in accordance with CVR and RE classiﬁcation. In the following sections, we review relevant CE and PT models according to model functionality and structure. A. Traﬃc ﬂow management models: Car Evacuation (CE). Cell Transmission Models (CTM). CTM (Daganzo, 1994) partition the car traﬃc network into grids and control the traﬃc ﬂow advancement over adjacent grids (cells). These models are obtained by transforming differential equations of traﬃc ﬂow into simpler difference equations resulting in an LP. A bi-level CTM is given by Liu, Lai, and Chang (2006) which maximizes traﬃc ﬂow and minimizes travel time. A combined optimization-simulation model is used to route the traﬃc ﬂow. Chiu and Zheng (2007) consider multi-group evacuation where some groups (e.g., nurses) have priority over others. Ben-Tal, Chung, Mandala, and Yao (2011) tackle dynamic traﬃc assignment problem such that traﬃc risks are minimized. The researchers use Robust Optimization to deal with traﬃc ﬂow uncertainty. Location–allocation models. Kongsomsaksakul, Yang, and Chen (2005) provide a bi-level location–allocation model for demand allocation to shelters and route selection. Both network links and shelters have capacity limits. The researchers use a Genetic Algorithm (GA) to minimize evacuation time and total travel time. MCNF models. Miller-Hooks and Patterson (2004) minimize network clearance time in a time dependent quickest ﬂow model (an LP) where travel times and link capacities are time-variant. Cova and Johnson (2003) discuss an MIP model for lane based evacuation routing where delays due to intersections and lane changes are considered. Solution method is based on successive shortest path algorithm.

59

A mixed integer capacitated maximum ﬂow model is proposed by Lim, Zangeneh, Baharnemati, and Assavapokee (2012) where the goal is to maximize number of evacuees. Dijkstra’s algorithm is used to ﬁnd paths and a greedy maximum ﬂow algorithm is used to determine the ﬂows. A multimodal NF model is described in Naghawi and Wolshon (2012) where routes are optimized by simulation. Campos, Bandeira, and Bandeira (2012) solve a maximum ﬂow problem with an iterative heuristic to identify two independent paths form each affected area to shelters. The idea is to minimize intersection delays and traﬃc congestion. Cui, An, and Zhao (2014) minimize the costs of evacuation and the contraﬂows of rescuers as well as lane reversal costs. They solve the nonlinear MCNF model using BARON (GAMS). B. Public Transit models (PT). Simulation models. Chen and Chou (2009) deal with the assignment of buses to pickup points in a nonotice evacuation. Simulation with contraﬂows is used to solve the problem so that traﬃc speed, system travel time, and network clearance time are minimized. CVR/LR/RE models. Murray-Tuite and Mahmassani (2003) minimize travel time including evacuee waiting time in a CVR model where capacitated vehicles (that can be anywhere in the network) pickup evacuees and transfer them to shelters. Margulis, Charosky, Fernandez, and Centeno (2006) propose a RE model that decides on the number of trips made between given pairs of pickup points and shelter nodes. Bus capacity cannot be exceeded on its trip and the goal is to maximize total number of evacuees. He, Zhang, Song, Wen, and Wu (2009) develop a stochastic LR model with demand uncertainty. Shelter selection, bus ﬂeet size, and bus routes are optimized using a combined metaheuristic (GA, hill climbing and Artiﬁcial Neural Networks). Song, He, and Zhang (2009) propose a LR model that selects pickup points among many in the city and optimize bus routes simultaneously in a no-notice evacuation. Evacuation time has a deadline and the number of evacuees is uncertain. The goal is to minimize total travel time and the researchers propose simulation based GA. Bish (2011) proposes both CVR and RE type models with multiple bus depots, split delivery and capacitated shelters in a short-notice evacuation. A route construction and improvement heuristic is implemented. Na, Xueyan, and Mingliang (2012) propose a RE model that minimizes maximum evacuation time and the costs of secondary evacuation that involves re-directing patients to other hospitals. The model assigns routes for capacitated buses, shelter capacities are also limited. Liu and Yu (2014, chap. 4) deﬁne a CVR model where two sets of decisions are made simultaneously: evacuees are optimally allocated to capacitated pickup points (based on walking distances from parking lots) and optimal bus routes are constructed. The goals are to minimize total travel distance and maximize the number of evacuees transported. DNF-TS models. Sayyady (2007) and Sayyady and Eksioglu (2010) represent waiting time of evacuees over a TS network in a no-notice evacuation. The model is solved by a Tabu Search that identiﬁes vehicle routes with the goal of minimizing total evacuation time, travel time, and waiting time. 2.2. The recovery phase In the recovery phase, planning issues are concerned with proper damage assessment, an effective reverse logistics system for debris disposal and recycling, and infrastructure rebuilding with minimum cost and duration. Table 3 summarizes model features for the recovery phase. Recovery models usually concentrate in two areas: road and other infrastructure restoration, and debris management that includes removal, disposal, and recycling. According to a FEMA report (2007), debris cleanup operations consist of two phases. During the ﬁrst phase, the goal is to clear debris from evacuation routes and other important paths to ensure traﬃc ﬂow in affected areas.

60

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

Table 3 Models for recovery phase. Citation

Objective(s)

Model type/features

Solution method

Fetter and Rakes (2012)

Debris facility setup and debris logistics cost

Exact

Hu and Sheu (2013)

Costs of transport, delay risk, psychological

Chang and Nojima (2001)

Measuring travel time extension due to blocked links in network Link criticality measure for network vulnerability Measuring areal network accessibility

Debris facility setup, location selection, transport (MIP) Multi-stage debris collection supply chain (LP) Network accessibility measure based on shortest path extensions Population based Chang and Nojima (2001) network accessibility index Measure based on average length of all paths between all pairs of nodes in area Grid based accessibility measure Bi-level LRM (CVR): repair due date, dozer travel times, limited no. of dozers

Sohn (2006) Allen, Liu, and Singer (1993) Jenelius and Mattsson (2012) Chen and Tzeng (1999, 2000)

Feng and Wang (2003) Özdamar et al (2014)

Tuzun Aksu and Özdamar (2014) Yan and Shih (2009) Maya Duque and Sörensen (2011) Chen et al. (2011) Liberatore et al. (2012) Matisziw and Murray (2009)

Measuring unmet travel demand 1) Minimize travel time, work time, idle time of work teams 2) Maximize traﬃc ﬂow Maximize repaired road length, number of repaired links, minimize risk of work teams Maximize network accessibility over repair time Maximize cumulative number of open paths over repair time Minimize maximum repair and delivery costs, demand satisfaction equity Minimize weighted shortest paths between pairs of regions Weighted relief delivery and unmet demand Minimize relief delivery time, unmet demand, maximize arc reliability Identify most critical p links (maximize ﬂow)

Exact, Pareto by constraint adding – Measuring link criticality by sequential link dropping on GIS map – GIS software Multi objective GA

LRM (CVR, multi-tour TSP): manpower, dozer clearance capacities Recursive LRM (parallel machine scheduling model): dozer work sequence with clearance times and travel times Dynamic PRM (MIP): repair sequence under resource constraints LRDM (DNF-TS, MIP)

Lingo-VRP package

Exact, 3 step heuristic

LRDM (binary MCNF): repair budget

GRASP metaheuristic

LRDM (CVR): non-split deliveries LRDM (binary MCNF): repair budget

GA Exact, lexicographic optimization

Path based network vulnerability model (BP): path aggregation constraints

Exact

Accessibility based link group criticality and makespan related heuristics Exact

Model types – LRM: Link based Repair Model, PRM: Path based Repair Model, LRDM: Link based Repair/Delivery Model, CVR, MCNF, DNF-TS, MIP, LP, BP deﬁned as in Tables 1 and 2.

This operation should start immediately after the disaster strikes and complete within 72 hours because road blockages impede rescuers, emergency services, and lifeline support (Kobayashi, 1995). In the second phase, all other debris are collected, reduced, transported, temporarily stored, recycled, and disposed. These operations could take months. Examples of the ﬁrst and second phases of post disaster debris management are given by Lauritzen (1998). In the next two sections, we summarize models for infrastructure restoration and subsequent debris management. 2.2.1. Models for infrastructure restoration Network accessibility measures. Road restoration activities involve clearing roadside debris and restoring the road network in order to open up evacuation routes and other important paths. A similar analogy can be made for the restoration of other types of infrastructure such as power and ﬁberoptic networks that share similar topologies. Hence, any infrastructure repair operation can be conducted eﬃciently by identifying the optimal order in which critical blocked links in the network are cleared. The goal would be to restore major paths as early as possible while maximizing network accessibility throughout the operation. Thus, network accessibility measures need to be deﬁned. Chang and Nojima (2001) deﬁne it as the length extension in post-disaster shortest paths between all pairs of locations in a road network. Sohn (2006) enhances this measure by weighing locations according to their populations and traﬃc densities. A road or link criticality index calculates the impact of its blockage on network accessibility. Sohn (2006) and Jenelius and Mattsson (2012) adopt a slow sequential approach that closes one link at a time to observe its impact on the whole network. This approach does not take the combined impacts of groups of blocked links into account; hence, the restoration plans based on sequential criticality index ordering might be suboptimal. Link and path based repair models. Link based repair models represent roads as arcs in the network and path based models represent them as nodes. A path is an ordered set of nodes. Path based models

are more compact than link based models and may converge faster because they explicitly consider links shared by multiple paths. These models involve path related goals leading to maximizing network accessibility or minimizing network vulnerability. On the other hand, link based models involve goals such as maximizing length of restored links, maximizing network ﬂow or minimizing the repair operation’s completion time (makespan). A. Link based repair models (LRM). CVR type. Chen and Tzeng (2000) propose a bi-level link based model where a due date is imposed for the completion of tasks, travel times between tasks are explicitly considered, and repair resources are unlimited. First level goals are to minimize travel, work, and idle times of work teams. Second level goal is to maximize traﬃc ﬂow. Chen and Tzeng (1999) use a multi-objective GA to solve this problem on a network with 24 nodes. Feng and Wang (2003) consider limitations on dozers, manpower, and debris clearing capacity. The goals considered are maximizing total length of accessible links, maximizing total number of restored links, and minimizing risk of work teams. The formulation involves a multi-tour TSP (each link is visited only once). The model is tested on a 50 node network with ten damaged links. Parallel machine scheduling type. Özdamar, Tuzun Aksu, and Ergunes (2014) aim to maximize the cumulative network accessibility measure deﬁned by Chang and Nojima (2001) over the road repair duration. The researchers propose heuristics that favor both repair operation makespan and network accessibility. Dozer clearance capacities, dozer travel times between tasks, and number of available dozers make up the model restrictions. B. Link based integrated repair and relief delivery models (LRDM). DNFTS type. Yan and Shih (2009) propose a link based DNF-TS model with work team trips and relief material ﬂows over repaired roads. An equity constraint exists for demand satisfaction. The goal is to minimize maximum repair and relief delivery costs. A three-step heuristic is proposed where blocked links are prioritized ﬁrst, worker schedules

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

and commodity ﬂows are optimized next. A 46 node network with 25 repair points is solved in 900 CPU seconds. CVR type. Chen, Peña-Mora, and Ouyang (2011) propose a nonsplit delivery CVR model to distribute relief over repaired roads. GA is used to optimize a weighted objective composed of maximizing relief delivery and minimizing unmet demand. MCNF type. A network ﬂow model is proposed by Liberatore, Ortuño, Tirado, Vitoriano, and Scaparra (2012) for relief distribution over a network with broken linkages. A budget constraint is imposed on ﬁxing links implicitly limiting the equipment and the number of work teams used in the operation. Several goals, such as delivery time, satisﬁed demand, and reliability of arcs are proposed. Maya Duque and Sörensen (2011) address a similar repair problem under budget constraints using a ﬁxed cost network ﬂow formulation for minimizing the cost of ﬂows from each rural center to the nearest regional center. C. Path based repair models (PRM). Network vulnerability models. A path based approach is adopted by Matisziw and Murray (2009) who identify a ﬁxed number of critical node and arc blockages that would prevent traﬃc ﬂow the most. The model addresses both mitigation and response phase planning (to protect most critical linkages in preparedness phase and recover them ﬁrst in response phase) and aims at maximizing ﬂows over broken source-sink paths. The most important feature of this model is the introduction of path aggregation constraints that dispose of the necessity to enumerate all source-sink paths. The researchers illustrate the mechanics of the model on a network of 23 nodes and 34 arcs. Network accessibility models. Tüzün Aksu and Özdamar (2014) propose a dynamic path based model that identiﬁes the optimal repair order of blocked links during a three-day work shift given a limited number of equipment. The goal is to open all access paths from each location in the network as early as possible and this maximizes the cumulative network accessibility throughout the restoration operation. Two real road network problems with up to 400 links (of which 70 are broken) are solved within 1 hour of computation time. 2.2.2. Models for debris management Brown, Milke, and Seville (2011) provide a wide review on post disaster waste treatment options. Different types of waste result from different disaster categories. While construction and demolition waste (C&D) results from all disaster types, sediments result from hurricanes, quakes and ﬂoods; green waste results from all but ﬁres; and ash results from quakes and ﬁres. The major contribution to debris comes from C&D. In the survey, the researchers emphasize the environmental, social, economic, and legal aspects that are the concern of the second phase in debris management. In particular, the speed needed in disposal due to health hazards, the requirement of accuracy in calculating disposal costs, psychological effects of allowing property owners to deconstruct, and re-use C&D waste (Denhart, 2009), vermin related public hazards (Petersen, 2004) are mentioned. The second phase debris management models usually concentrate on locating temporary storage and recycling facilities around the affected area as well as determine debris transportation routes. Supply chain models. Fetter and Rakes (2012) present a facility location model for locating temporary disposal and storage reduction facilities to support debris clean-up operations. The objective is to minimize the costs of debris collection, recycling, and disposal as well as the ﬁxed costs of enabling recycling technology at temporary facilities. Hu and Sheu (2013) propose a reverse logistics chain for debris disposal with four stages: on site processing and recycling, nearby temporary site processing and recycling, mass processing by permanent facilities, disposal and recycling, and reproduction for construction products (bricks). A network model that enables material balance at each stage is developed and the goals are to minimize logistics costs, minimize risk based penalty due to delay of debris removal,

61

and minimize psychological costs of delay. The researchers analyze the Wenchuan quake (2008 China) as a case study. They observe that risk induced costs are decreased by 37 percent at the expense of 243 percent of an increase in logistics costs. 3. Use of information systems in humanitarian logistics It is very diﬃcult, if not impossible, to implement solution procedures in real life without a proper information system that bundles mathematical models in a user-friendly interface. Moreover, information systems have the potential to act as a liaison between practitioners and academic researchers in humanitarian logistics as well as among different disaster lifecycle stages. For a literature review for the use of information systems in humanitarian logistics and future research directions, Sangiamkul and van Hillegersberg (2011) and Ortuño et al. (2013) can be referred. Blecken and Hellingrath (2008) compare ﬁve practitioner software (i.e., SUMA, LSS, HELIOS (HELIOS, 2013), LogistiX, Sahana) and conclude that critical decisions such as inventory management, supply planning, routing and scheduling are not addressed by any of them. In addition to the ﬁve software discussed in Blecken and Hellingrath (2008), Ortuño et al. (2013) reR , view Disaster Management Information System (DMIS), FleetWave and Humanitarian Free and Open Source Software (HFOSS) (Tucker, Morelli, & de Lanerolle, 2011) as well. Ortuño et al. (2013) also report the humanitarian relief efforts where the software was used. Howden (2009) explains where information systems can play a role from a disaster operations management lifecycle perspective. Chib and Komathi (2009) extend the framework for technology-communitymanagement model to assess vulnerability in Asian countries for disaster recovery. Here, the use of different types of information systems in humanitarian logistics is addressed for the response and recovery phases. In Table 4, we provide academic studies and applications that correspond to different technologies. 3.1. Maps and geographic information systems Similar to the scarcity of relief supplies and equipment; ﬁeld data for the location and impact of the disaster, the availability of the transportation infrastructure, and estimated demand quantity are scarce resources in the aftermath of disasters. Maps and Geographic Information Systems (GIS) are vital to provide these ﬁeld data. GIS research on disaster management, GIS applications, and potential future study topics are listed in Cutter (2003). Disaster risk maps and alerts for upcoming disasters are developed to assist relief operations. A good example is the Global Disaster Alert and Coordination System (GDACS) of United Nations and European Commission (GDACS, 2014 ). GDACS analyses GIS data, maps, Satellite Mapping Coordination System, ﬁeld data (e.g., reports, photos/videos, Geo-Pictures), mass and social media information and presents the estimate of the disaster impact, risk, and magnitude via a global map. GDACS is now accessible by a smart phone app called iGDACS. Several researchers have integrated their models with maps and GIS. Chang et al. (2007) use a scenario planning approach for rescue center location problem in conjunction with GIS software for ﬂood emergencies. Saadatseresht, Mansourian, and Taleai (2009) propose an evolutionary approach using GIS software for solving a spatial multi-objective evacuation problem. Mete and Zabinsky (2010) address prepositioning of medical supplies using a simulation and visualization environment called RimSim (Campbell, Mete, Furness, Weghorst, & Zabinsky, 2008). Widener and Horner (2011) develop a model for deciding the size and location of aid distribution facilities for post-hurricane settings. Their model utilizes GIS and spatial optimization technologies to solve p-median and capacitated median type problems. An application of GIS based techniques for needs assessment is provided in Benini, Conley, Dittemore, and Waksman (2009). The

62

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

Table 4 Summary of the use of information systems in humanitarian logistics. Citation

Maps and/or GIS

Coordination network

Smart phone apps

GDACS (2014) Chang et al. (2007) Saadatseresht et al. (2009) Mete and Zabinsky (2010)

GIS data, map GIS GIS N/A

Mass and social media N/A N/A N/A

iGDACS N/A N/A N/A

Widener and Horner (2011) Benini et al. (2009) ACAPS (2014) MapAction (2014) SandyCoworking (2013) Humanitarian OpenStreetMap (2014) Chen et al. (2011) Zografos and Androutsopoulos (2008) Jotshi et al. (2009) Demir and Özdamar (2013) Huang et al. (2013) Pasupathy and Madina-Borja (2008) Vitoriano et al. (2009) Tomaszewski et al. (2006)

GIS GIS algorithms Heat maps GIS, GMES satellite imagery, QGIS GoogleMaps GIS ArcGIS GIS ArcGIS ArcGIS OpenStreetMap N/A Logistics maps Open source GIS

Özgüven and Özbay (2013) Yang et al. (2011)

N/A N/A

N/A UN joint logistics cluster Community of practice UN assessment and coordination team List of available spaces Web portal Collaborative GIS N/A N/A N/A Urban search and rescue American Red Cross Web based tool Geo-collaborative web portal N/A N/A

Assessment Capacities Project (ACAPS) is one example dedicated to better deal with the needs assessment operations in the response phase (ACAPS, 2014). Software such as MapAction depicts the needs and speciﬁc locations around the disaster area (MapAction, 2014). Crowd sourcing is a tool that was used during the Hurricane Sandy (SandyCoworking, 2013). This specialized crowd map helped businesses and individuals to move to safer places by populating information about available spots around the area. Humanitarian OpenStreetMap initiative provides key geographic data to help humanitarian responders. It is built by volunteers recording GPS, aerial imagery, and geographic data after a catastrophic event (Humanitarian OpenStreetMap, 2014). 3.2. Integrated systems Information systems are used in a variety of application areas such as damage and needs assessment, locating aid distribution facilities, and vehicle routing in the response and recovery phases. Unfortunately, not many studies can be reported that integrate available technologies with proper mathematical models through several disaster lifecycle stages. A comprehensive study for the response phase is given in Chen et al. (2011) which addresses the challenge of construction equipment allocation, management, and routing for urban search and rescue operations under scarce resource disaster relief environment. The researchers combine three technologies (i.e., GIS, resource tracking and management, and decision making by a mathematical model) in a framework with mobile resource requesting ability. Zografos and Androutsopoulos (2008) develop an integrated decision support system (DSS) for hazardous material transportation and emergency response decisions. The implemented DSS has logistics, crisis management, database, GIS, and human machine interface subsystems. Jotshi, Gong, and Batta (2009) use data fusion to dispatch and route emergency vehicles with the aim of minimizing the response time; hence, maximizing the life expectancy. Data fusion technique is used in conjunction with ArcGIS, Tele-Atlas, and HAZUS technologies. Demir and Özdamar (2013) describe a hierarchical relief delivery and casualty pickup coordination system using dynamic data transferred from ArcGIS database. The system provides on-line user interface supporting access to the database from Android based hand devices. Huang et al. (2013) integrated visualization of real time crowd-

Other technologies

Satellite, geo-pictures N/A Spatial multi-objective optimization RimSim (a simulation and visualization environment) N/A Spatial optimization N/A NASA satellite imagery NOMAD N/A N/A GoogleEarth, portable satellite data modem, hand held GPS units N/A Ushahidi platform Available Aerial imagery Mobile resource requesting ability Resource tracking and management N/A Database, human machine interface N/A Data fusion, Tele-Atlas, and HAZUS Android app On-line user interface SMS Matlab, ArcGIS, Python N/A Excel, Access, VBA, SQL N/A Java, Access, GAMS PDA or tablet MapBuilder API and JetSpeed Portal N/A N/A

RFID RFID

sourcing data on OpenStreetMap with SMS messages for demand estimation. Huang et al. (2013) implemented a routing heuristic for urban search and rescue teams. An example of integrated software packages applied for performance management in the American Red Cross (ARC) is explained by Pasupathy and Madina-Borja (2008). Equipped with a Data Envelopment Analysis model, this tool was implemented to measure the performances of about 1000 ARC chapters with respect to criteria such as ﬁnancial resources, capacity creation, service delivery, and customer effectiveness. Vitoriano et al. (2009) introduce a multiobjective model with integration of Java, GAMS, and Microsoft Access to support transport operations in disaster relief. DMIS (World Disaster Report, 2013, p. 123) is an integrated platform that provides real time data about disasters and tools for operations of International Federation of Red Cross and Red Crescent (IFRC). This web based tool is only available for IFRC staff. Although not develR (FleetWave, oped speciﬁcally for humanitarian logistics, FleetWave 2014) is a web-based enterprise ﬂeet management software used by large humanitarian organizations such as IFRC and World Food Programme (WFP) (Ortuño et al., 2013). 3.3. Coordination networks Coordination is one of the biggest challenges in humanitarian logistics (Balçık, Beamon, Krejci, Muramatsu, & Ramirez, 2010). Several stakeholders compete for scarce resources in a chaotic environment. Information systems provide a unique opportunity to overcome coordination challenges. There are speciﬁc tools for coordination some of which were discussed in Section 3.1. Bartell, Lappenbusch, Kemp, and Haselkorn (2006) give a review about the use of information systems in effective coordination and communication Tomaszewski, MacEachren, Pezanowski, Liu, and Turton (2006) present an example with a geo-collaborative web portal to address the coordination and collaboration challenges in a disaster relief environment. The researchers use open source software such as MapBuilder API and JetSpeed Portal. Open source software community took responsibility to help humanitarian logisticians in the ﬁeld. Humanitarian Free and Open Source Software (HFOSS) project has been working on both engaging undergrad students in FOSS community as well as doing good with good computer science practices. Portable Open Source

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

Information Tool (POSIT) and Collabit are two successful tools developed by HFOSS project contributors (Ortuño et al., 2013; Tucker et al., 2011). In addition to speciﬁc tools, there exist non-governmental organizations that aim at increasing the eﬃciency of disaster response and recovery by using Information and Communication Technology at a greater scale (Aidmatrix, 2014; NetHope, 2013). The Aidmatrix R Network for Humanitarian Relief provides comprehensive supply chain management software including modules such as planning, procurement, transportation, warehouse, online ordering, distribution, ﬂeet management, asset management, and distribution. The International Network of Crisis Mappers (Crisis Mappers Net, 2014) is a network of various stakeholders working at the crossing of disasters, novel technologies, crowd-sourcing, and crisis mapping. This network has more than 7000 members in over 160 countries, who are aﬃliated with over 3000 different institutions from academic researchers to actual practitioners on the ﬁeld. Logistics Cluster approach is also a successful example in coordinating scarce logistical resources in a disaster area. The information ﬂow among humanitarian organizations is managed and up-to-date maps about the available warehouses, trucks, etc. around the disaster area are released through a logistics cluster (Tomasini & Van Wassenhove, 2005). 3.4. Other technologies Remote sensing techniques such as satellite or camera image processing are used for damage (Pesaresi, Gerhardinger, & Haag, 2007) and needs assessment (Tatham, 2009). An example for using an unmanned aerial vehicle system for needs assessment is illustrated in Tatham (2009). Radio Frequency Identiﬁcation (RFID) sensors are also investigated for their use in humanitarian logistics settings. Yang, Yang, and Yang (2011) design a sensor network with a hybrid implementation of active and passive RFID tags for humanitarian logistics center management. Özgüven and Özbay (2013) discuss both off-line and on-line use of RFID information to manage inventory for humanitarian relief efforts. From the discussions above, we observe that disaster management software cover a wide portfolio of dispersed applications; however, holistic software that integrate all phases of the disaster lifecycle are yet to be developed. Moreover, some software those seem to offer an integrated solution are not freely available to humanitarian aid workers all around the world. 4. Conclusion According to the claims of climate scientists and the trend in disaster frequency during the recent decades, during the 21st century, residents of the earth will suffer severely from the environmental outcomes of the 20th century’s technological advances (IPCC Report, 2014). Even the most optimistic views on the progress of global warming have demoralizing aspects in terms of clean water availability, temperature rise, sea level changes, droughts, ﬂoods, wild ﬁres, and earthquakes. Global awareness of the forthcoming calamities has risen due to extensive media coverage of climate change and of the high impact disasters that took place in the last decade. Academicians have duly responded by developing mathematical models and solution algorithms that deal with different aspects of disasters. Information technologies such as GIS and remote sensing entered the ﬁeld of disaster management along with mobile hand devices that enable access to/from people on the street. The detailed discussions held here on the modeling aspects of the literature implicate that though preparedness plans can be zone based and rough, response and recovery plans have to include street address level of detail in order to construct routes for relief delivery, repair equipment delivery, detailed itineraries for rescuers, etc. Unfortunately, algorithms for commercial routing problems do not satisfy

63

the needs of humanitarian logistics; therefore, methods that deal with large scale disasters are not readily available. Though the proposed logistics planning models have good practical features, most of the models’ computational burdens might prevent them from being used in actual disasters. From practical and academic perspectives, on-line algorithms that aim at both service quality and optimal resource utilization would be most welcome in high impact disasters. Another area of improvement would be developing more solvable models that treat all aspects of recovery in an integrated manner including debris transportation, road repair, relief delivery, and evacuation during the ﬁrst phase of recovery. Although integrated models exist, the models are either oversimpliﬁed or too complex and unsolvable. Proposed metaheuristics are not yet tested on realistic size networks. As discussed in Section 3, the developed enabling technologies that should make solution methods available to aid practitioners are quite dispersed in their functionality and accessibility. Although there have been several attempts in recent years to integrate some type of information systems into the humanitarian logistics models, the use of technology is still not prevalent in humanitarian logistics. Practitioner software lack sophisticated mathematical models, hence optimal results, but have nice graphical user interfaces and fast solutions. On the other hand, few academic researchers attempt to integrate their mathematical models into a decision support system equipped with GIS, maps, real-time data, and a user-friendly interface. Although it is not very easy to combine the best values of practitioners and academic researchers, the use of technology is a must in order to achieve success in humanitarian logistics operations. A holistic approach such as webbased Enterprise Resource Planning (ERP) systems is yet to be seen in this area. The commercial versions of such systems are readily available but they need to be adapted to disaster context. Unfortunately, most of the disaster response costs are born by governments and private damage costs by insurance companies. Unless disasters use up country GDPs more and more, government funding will probably not be made available to build a disaster-ERP, i.e., Disaster Resource Planning (DRP) software. The ﬁrst step is to calculate governmental disaster costs correctly and conduct a gap analysis to validate building the DRP. In order to optimize response phase resource utilization, the DRP software and database should be globally accessible. International cooperation should be promoted to design a global disaster supply chain managed by the DRP, setting political differences aside. The challenges that lie ahead of a successful implementation is lack of real-time data, fragmented nature of humanitarian organizations, lack of commonly accepted interoperability standards among different humanitarian organizations, and not realizing the importance to balance good bookkeeping practices and to act fast in emergency relief operations. These barriers are rather concerned with the human way of doing things than with the lack of available technologies. We believe that they can be overcome with adequate determination. Acknowledgments This work was partially supported by The Scientiﬁc and Technological Research Council of Turkey (TUBITAK) under grant numbers 110M578 and 113M493. The authors would also like to thank the anonymous referees for their helpful comments and suggested improvements. References ACAPS, (2014). Assessment Capacities Project (ACAPS). Available online at: http://www.acaps.org/en/pages/what-is-acaps (accessed 09 November 2014). Afshar, A., & Haghani, A. (2012). Modeling integrated supply chain logistics in realtime large-scale disaster relief operations. Socio-Economic Planning Sciences, 46(4), 327–338. Aidmatrix, (2014). Available online at: http://www.aidmatrix.org/ (accessed 09 November 2014).

64

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65

Allen, W. B., Liu, D., & Singer, S. (1993). Accesibility measures of US metropolitan areas. Transportation Research Part B: Methodological, 27(6), 439–449. Altay, N., & Green, W. G., III (2006). OR/MS research in disaster operations management. European Journal of Operational Research, 175(1), 475–493. Anaya-Arenas, A. M, Renaud, J., & Ruiz, A. (2014). Relief distribution networks: A systematic review. Annals of Operations Research, 1–27. Apte, A. (2009). Humanitarian logistics: A new ﬁeld of research and action. Foundations and Trends in Technology, Information and Operations Management, 3(1), 1–100. Balçık, B., & Beamon, B. M. (2008). Facility location in humanitarian relief. International Journal of Logistics, 11(2), 101–121. Balçık, B., Beamon, B. M., Krejci, C. C., Muramatsu, K. M., & Ramirez, M. (2010). Coordination in humanitarian relief chains: Practices, challenges and opportunities. International Journal of Production Economics, 126(1), 22–34. Balçık, B., Beamon, B. M., & Smilowitz, K. (2008). Last mile distribution in humanitarian relief. Journal of Intelligent Transportation Systems, 12(2), 51–63. Barbaroso˘glu, G., & Arda, Y. (2004). A two-stage stochastic programming framework for transportation planning in disaster response. Journal of the Operational Research Society, 55(1), 43–53. Barbaroso˘glu, G., Özdamar, L., & Cevik, A. (2002). An interactive approach for hierarchical analysis of helicopter logistics in disaster relief operations. European Journal of Operational Research, 140(1), 118–133. Bartell, A. L., Lappenbusch, S., Kemp, R. B., & Haselkorn, M. (2006, October). Improving humanitarian relief information and communication systems through research (pp. 156–162), IEEE. Benini, A., Conley, C., Dittemore, B., & Waksman, Z. (2009). Survivor needs or logistical convenience? Factors shaping decisions to deliver relief to earthquake-affected communities, Pakistan 2005–06. Disasters, 33(1), 110–131. Ben-Tal, A., Chung, B. D., Mandala, S. R., & Yao, T. (2011). Robust optimization for emergency logistics planning: Risk mitigation in humanitarian relief supply chains. Transportation Research Part B: Methodological, 45(8), 1177–1189. Berkoune, D., Renaud, J., Rekik, M., & Ruiz, A. (2012). Transportation in disaster response operations. Socio-Economic Planning Sciences, 46(1), 23–32. Bish, D. R. (2011). Planning for a bus-based evacuation. OR Spectrum, 33(3), 629–654. Blecken, A., & Hellingrath, B. (2008). Supply chain management software for humanitarian operations: Review and assessment of current tools (pp. 342–351). Brown, C., Milke, M., & Seville, E. (2011). Disaster waste management: A review article. Waste Management, 31(6), 1085–1098. Campbell, A. M., Vandenbussche, D., & Hermann, W. (2008). Routing for relief efforts. Transportation Science, 42(2), 127–145. Campbell, B. D., Mete, H. O., Furness, T., Weghorst, S., & Zabinsky, Z. (2008). Emergency response planning and training through interactive simulation and visualization with decision support (pp. 176–180), IEEE. Campos, V., Bandeira, R., & Bandeira, A. (2012). A method for evacuation route planning in disaster situations. Procedia-Social and Behavioral Sciences, 54, 503–512. Caunhye, A. M., Nie, X., & Pokharel, S. (2012). Optimization models in emergency logistics: A literature review. Socio-Economic Planning Sciences, 46(1), 4–13. Chang, M. S., Tseng, Y. L., & Chen, J. W. (2007). A scenario planning approach for the ﬂood emergency logistics preparation problem under uncertainty. Transportation Research Part E: Logistics and Transportation Review, 43(6), 737–754. Chang, S. E., & Nojima, N. (2001). Measuring post-disaster transportation system performance: The 1995 Kobe earthquake in comparative perspective. Transportation Research Part A: Policy and Practice, 35(6), 475–494. Chen, A. Y., Peña-Mora, F., & Ouyang, Y. (2011). A collaborative GIS framework to support equipment distribution for civil engineering disaster response operations. Automation in Construction, 20(5), 637–648. Chen, C. C. F., & Chou, C. S. (2009). Modeling and performance assessment of a transitbased evacuation plan within a contraﬂow simulation environment. Transportation Research Record: Journal of the Transportation Research Board, 2091(1), 40–50. Chen, Y. W., & Tzeng, G. H. (1999). A fuzzy multi-objective model for reconstructing the post-quake road-network by genetic algorithm. International Journal of Fuzzy Systems, 1(2), 85–95. Chen, Y. W., & Tzeng, G. H. (2000, July). Determining the optimal reconstruction priority of a post-quake road network. Computing in civil and building engineering (2000), 686–693 ASCE. Chib, A., & Komathi, A. L. E. (2009). Extending the technology-community-management model to disaster recovery: Assessing vulnerability in rural Asia (pp. 328–336), IEEE. Chiu, Y. C., & Zheng, H. (2007). Real-time mobilization decisions for multi-priority emergency response resources and evacuation groups: Model formulation and solution. Transportation Research Part E: Logistics and Transportation Review, 43(6), 710–736. Cova, T., & Johnson, J. (2003). A network ﬂow model for lane-based evacuation routing. Transportation Research Part A: Policy and Practice, 37(7), 579–604. Crisis Mappers Net, (2014). Available online at: http://crisismappers.net/ (accessed 09 November 2014). Cui, J., An, S., & Zhao, M. (2014). A generalized minimum cost ﬂow model for multiple emergency ﬂow routing. Mathematical Problems in Engineering, 2014 http://dx.doi.org/10.1155/2014/832053 . Cutter, S. L. (2003). GI science, disasters, and emergency management. Transactions in GIS, 7(4), 439–446. Daganzo, C. F. (1994). The cell transmission model: A dynamic representation of highway traﬃc consistent with the hydrodynamic theory. Transportation Research Part B: Methodological, 28(4), 269–287. de Angelis, V., Mecoli, M., Nikoi, C., & Storchi, G. (2007). Multiperiod integrated routing and scheduling of World Food Programme cargo planes in Angola. Computers & Operations Research, 34(6), 1601–1615.

de la Torre, L. E., Dolinskaya, I. S., & Smilowitz, K. R. (2012). Disaster relief routing: Integrating research and practice. Socio-economic Planning Sciences, 46(1), 88–97. Demir, O., & Özdamar, L. (2013). A GIS-based emergency response system for disaster relief. Denhart, H. (2009). Deconstructing disaster: Psycho-social impact of building deconstruction in Post-Katrina New Orleans. Cities, 26(4), 195–201. Duran, S., Gutierrez, M. A., & Keskinocak, P. (2011). Pre-positioning of emergency items for CARE international. Interfaces, 41(3), 223–237. FEMA. (2007). Public Assistance; Debris Management Guide. Feng, C. M., & Wang, T. C. (2003). Highway emergency rehabilitation scheduling in post-earthquake 72 hours. Journal of the 5th Eastern Asia Society for Transportation Studies, 5, 3276–3285. Fetter, G., & Rakes, T. (2012). Incorporating recycling into post-disaster debris disposal. Socio-Economic Planning Sciences, 46(1), 14–22. FleetWave, (2014). Web-based ﬂeet management software. Available online at: http:// www.chevinﬂeet.com/GB/ﬂeetwave_web_based_enterprise_ﬂeet_management_ solution.asp (accessed 09 November 2014). Galindo, G., & Batta, R. (2013). Review of recent developments in OR/MS research in disaster operations management. European Journal of Operational Research, 230(2), 201–211. GDACS. (2014). Global Disaster Alert and Coordination System. Available online at: http://www.gdacs.org/ (accessed 09 November 2014). Gu, Y. (2011). Research on optimization of relief supplies distribution aimed to minimize disaster losses. Journal of Computers, 6(3), 603–609. Guha-Sapir, D., Hoyois, P., & Below, R. (2014). Annual disaster statistical review 2013: The numbers and trends. Brussels: CRED. Haghani, A., & Oh, S. C. (1996). Formulation and solution of a multi-commodity, multimodal network ﬂow model for disaster relief operations. Transportation Research Part A: Policy and Practice, 30(3), 231–250. He, S., Zhang, L., Song, R., Wen, Y., & Wu, D. (2009). Optimal transit routing problem for emergency evacuations. Helios. (2013). Helios Overview. Available online at: http://www.fritzinstitute. org/prgTech-HELIOS_Overview.htm (accessed 29 November 2013). Hobeika, A. G., & Kim, C. (1998). Comparison of traﬃc assignments in evacuation modeling. Engineering Management, IEEE Transactions on, 45(2), 192–198. Howden, M. (2009, May). How humanitarian logistics information systems can improve humanitarian supply chains: A view from the ﬁeld. Gothenburg, Sweden: Prentice Hall. Hsueh, C. F., Chen, H. K., & Chou, H. W. (2008). Dynamic vehicle routing for relief logistics in natural disasters. In T. Caric, & H. Gold (Eds.), Vehicle routing problem (pp. 71–84). Vienna, Austria: InTech. Hu, Z. H., & Sheu, J. B. (2013). Post-disaster debris reverse logistics management under psychological cost minimization. Transportation Research Part B: Methodological, 55, 118–141. Huang, A., Ma, A., Schmidt, S., Xu, N., Zhang, B., Meineke, L., et al. (2013). Integration of Real Time Data in Urban Search and Rescue. Center for the Commercialization of Innovative Transportation Technology, Transportation Center, Northwestern University. Technical report. Huang, M., Smilowitz, K., & Balçık, B. (2012). Models for relief routing: Equity, eﬃciency and eﬃcacy. Transportation Research Part E: Logistics and Transportation Review, 48(1), 2–18. Humanitarian OpenStreenMap, (2014). Available online at: http://hot.openstreet map.org/about (accessed 09 November 2014). Ichoua, S. (2010). Humanitarian logistics network design for an effective disaster response. IPCC Report 2014: Climate change 2014: Impacts, adaptation and vulnerability. www.ipcc.ch/report/ar5/wg2 (accessed 11 August 2014). Jenelius, E., & Mattsson, L. G. (2012). Road network vulnerability analysis of areacovering disruptions: A grid-based approach with case study. Transportation Research Part A: Policy and Practice, 46(5), 746–760. Jotshi, A., Gong, Q., & Batta, R. (2009). Dispatching and routing of emergency vehicles in disaster mitigation using data fusion. Socio-Economic Planning Sciences, 43(1), 1–24. Kobayashi, Y. (1995, June). Disasters and the problems of wastes-institutions in Japan and issues raised by the Great Hanshin-Awaji earthquake. Kongsomsaksakul, S., Yang, C., & Chen, A. (2005). Shelter location-allocation model for ﬂood evacuation planning. Journal of the Eastern Asia Society for Transportation Studies, 6(1), 4237–4252. Lauritzen, E. K. (1998). Emergency construction waste management. Safety Science, 30(1), 45–53. Liberatore, F., Ortuño, M. T., Tirado, G., Vitoriano, B., & Scaparra, M. P. (2012). A hierarchical compromise model for the joint optimization of recovery operations and distribution of emergency goods in Humanitarian Logistics. Computers & Operations Research http://dx.doi.org/10.1016/j.cor.2012.03.019 . Lim, G. J., Zangeneh, S., Baharnemati, M. R., & Assavapokee, T. (2012). A capacitated network ﬂow optimization approach for short notice evacuation planning. European Journal of Operational Research, 223(1), 234–245. Lin, Y. H., Batta, R., Rogerson, P. A., Blatt, A., & Flanigan, M. (2011). A logistics model for emergency supply of critical items in the aftermath of a disaster. Socio-Economic Planning Sciences, 45(4), 132–145. Liu, Y., Lai, X. R., & Chang, G. L. (2006). Two-level integrated optimization system for planning emergency evacuation. Journal of Transportation Engineering, 132, 800–807. Liu, Y., Lai, X., & Chang, G. (2006). Cell-based network optimization model for staged evacuation planning under emergencies. Journal of the Transportation Research Board, 1964, 127–135.

L. Özdamar, M. A. Ertem / European Journal of Operational Research 244 (2015) 55–65 Liu, Y., & Yu, J. (2014). Emergency evacuation planning for highly populated urban zone: A transit based solution and optimal operational strategies. Emergency management. doi:10. 5772/33653. MapAction. (2014). Available online at: http://www.mapaction.org/about.html (accessed 09 November 2014). Margulis, L., Charosky, P., Fernandez, J., & Centeno, M. A. (2006, June). Hurricane evacuation decision-support model for bus dispatch. Mastrogiannidou, C., Boile, M., Golias, M., Theofanis, S., & Ziliaskopoulos, A (2009). Using transit to evacuate facilities in urban areas: A micro-simulation based integrated tool. Transportation Research Record: Journal of the Transportation Research Board 3439 (pp. 11–15). Matisziw, T. C., & Murray, A. T. (2009). Modeling s–t path availability to support disaster vulnerability assessment of network infrastructure. Computers & Operations Research, 36(1), 16–26. Maya Duque, P., & Sörensen, K. (2011). A GRASP metaheuristic to improve accessibility after a disaster. OR Spectrum, 33(3), 525–542. Mete, H. O., & Zabinsky, Z. B. (2010). Stochastic optimization of medical supply location and distribution in disaster management. International Journal of Production Economics, 126(1), 76–84. Miller-Hooks, E., & Patterson, S. S. (2004). On solving quickest time problems in timedependent, dynamic networks. Journal of Mathematical Modelling and Algorithms, 3(1), 39–71. Murray-Tuite, P. M., & Mahmassani, H. S. (2003). Model of household trip-chain sequencing in emergency evacuation. Transportation Research Record: Journal of the Transportation Research Board, 1831(1), 21–29. Na, L., Xueyan, S., & Mingliang, Q. (2012). A bi-objective evacuation routing engineering model with secondary evacuation expected costs. Systems Engineering Procedia, 5, 1–7. Naghawi, H., & Wolshon, B. (2012). Performance of traﬃc networks during multimodal evacuations: Simulation-based assessment. Natural Hazards Review, 13(3), 196–204. Najaﬁ, M., Eshghi, K., & Dullaert, W. (2013). A multi-objective robust optimization model for logistics planning in the earthquake response phase. Transportation Research Part E: Logistics and Transportation Review, 49(1), 217–249. Natarajarathinam, M., Capar, I., & Narayanan, A. (2009). Managing supply chains in times of crisis: A review of literature and insights. International Journal of Physical Distribution & Logistics Management, 39(7), 535–573. NetHope. (2013). Available online at: http://nethope.org/programs/emergencyresponse/ (accessed 29 November 2013). Nolz, P. C., Semet, F., & Doerner, K. F. (2011). Risk approaches for delivering disaster relief supplies. OR Spectrum, 33(3), 543–569. Ortuño, M. T., Cristóbal, P., Ferrer, J. M., Martín-Campo, F. J., Muñoz, S., Tirado, G., et al. (2013). Decision aid models and systems for humanitarian logistics. A survey. In Decision aid models for disaster management and emergencies (pp. 17–44). Atlantis Press. Özdamar, L. (2011). Planning helicopter logistics in disaster relief. OR Spectrum, 33(3), 655–672. Özdamar, L. (2013). Models and solutions in emergency logistics, invited keynote speech. Özdamar, L., & Demir, O. (2012). A hierarchical clustering and routing procedure for large scale disaster relief logistics planning. Transportation Research Part E: Logistics and Transportation Review, 48(3), 591–602. Özdamar, L., Demir, O., Bakır, E., & Yılmaz, S. (2013). Hierarchical optimization for helicopter mission planning in large-scale emergencies. In A. B. Badiru, & L. Racz (Eds.), Handbook of emergency response: A human factors and systems engineering approach (pp. 151–174) CRC Press, Taylor and Francis. Özdamar, L., Ekinci, E., & Küçükyazici, B. (2004). Emergency logistics planning in natural disasters. Annals of Operations Research, 129(1-4), 217–245. Özgüven, E. E., & Özbay, K. (2013). A secure and eﬃcient inventory management system for disasters. Transportation Research Part C: Emerging Technologies, 29, 171–196. Özdamar, L., & Pedamallu, C. S. (2011). A comparison of two mathematical models for earthquake relief logistics. International Journal of Logistics Systems and Management, 10(3), 361–373. Özdamar, L., Tuzun Aksu, D., & Ergunes, B. (2014). Coordinating debris cleanup operations in post disaster road networks. Socio-Economic Planning Sciences, 48, 249–262. Özdamar, L., & Yi, W. (2008). Greedy neighborhood search for disaster relief and evacuation logistics. Intelligent Systems, IEEE, 23(1), 14–23. Pasupathy, K. S., & Medina-Borja, A. (2008). Integrating Excel, Access, and Visual Basic to deploy performance measurement and evaluation at the American Red Cross. Interfaces, 38(4), 324–337. Perkins, J.A., Dabipi, I.K., & Han, L.D. (2001). Modeling transit issues unique to hurricane evacuations: North Carolina’s small urban and rural areas (No. 00924084 Final Report). Pesaresi, M., Gerhardinger, A., & Haag, F. (2007). Rapid damage assessment of built-up structures using VHR satellite data in tsunami-affected areas. International Journal of Remote Sensing, 28(13-14), 3013–3036.

65

Petersen, M. (2004, April). Restoring waste management following disasters. In International conference on post disaster reconstruction (pp. 22–23). Saadatseresht, M., Mansourian, A., & Taleai, M. (2009). Evacuation planning using multiobjective evolutionary optimization approach. European Journal of Operational Research, 198(1), 305–314. SandyCoworking. 2013. Available online at: https://sandycoworking.crowdmap.com/ (accessed 29 November 2013). Sangiamkul, E., & Hillegersberg, J. (2011). Research directions in information systems for humanitarian logistics: vol. 1 Lisbon. Sayyady, F. (2007). Optimizing the use of public transit system in no-notice evacuations in urban areas (Doctoral dissertation, Mississippi State University). Sayyady, F., & Eksioglu, S. D. (2010). Optimizing the use of public transit system during no-notice evacuation of urban areas. Computers & Industrial Engineering, 59(4), 488–495. Shen, Z., Dessouky, M. M., & Ordóñez, F. (2009). A two-stage vehicle routing model for large-scale bioterrorism emergencies. Networks, 54(4), 255–269. Sheu, J. B. (2010). Dynamic relief-demand management for emergency logistics operations under large-scale disasters. Transportation Research Part E: Logistics and Transportation Review, 46(1), 1–17. Sohn, J. (2006). Evaluating the signiﬁcance of highway network links under the ﬂood damage: An accessibility approach. Transportation Research Part A: Policy and Practice, 40(6), 491–506. Song, R., He, S., & Zhang, L. (2009). Optimum transit operations during the emergency evacuations. Journal of Transportation Systems Engineering and Information Technology, 9(6), 154–160. Tatham, P. (2009). An investigation into the suitability of the use of unmanned aerial vehicle systems (UAVS) to support the initial needs assessment process in rapid onset humanitarian disasters. International Journal of Risk Assessment and Management, 13(1), 60–78. Tomasini, R., & Van Wassenhove, L. (2005). Managing information in humanitarian crisis: UNJLC Website. INSEAD Case Study (5278). Tomaszewski, B. M., MacEachren, A. M., Pezanowski, S., Liu, X., & Turton, I. (2006, May). Supporting humanitarian relief logistics operations through online geocollaborative knowledge management (pp. 358–359) Digital Government Society of North America. Tucker, A., Morelli, R., & de Lanerolle, T. (2011, October). The Humanitarian FOSS Project: Goals, activities, and outcomes (pp. 98–101), IEEE. Tuzun Aksu, D., & Özdamar, L. (2014). A mathematical model for post-disaster road restoration: Enabling accessibility and evacuation. Transportation Research Part E: Logistics and Transportation Review, 61, 56–67. Tzeng, G. H., Cheng, H. J., & Huang, T. D. (2007). Multi-objective optimal planning for designing relief delivery systems. Transportation Research Part E: Logistics and Transportation Review, 43(6), 673–686. Ukkusuri, S. V., & Yushimito, W. F. (2008). Location routing approach for the humanitarian prepositioning problem. Transportation Research Record: Journal of the Transportation Research Board, 2089(1), 18–25. Vitoriano, B., Ortuño, M. T., Tirado, G., & Montero, J. (2011). A multi-criteria optimization model for humanitarian aid distribution. Journal of Global Optimization, 51(2), 189–208. Vitoriano, B., Ortuno, T., & Tirado, G. (2009). HADS, a goal programming-based humanitarian aid distribution system. Journal of Multi-Criteria Decision Analysis, 16(1-2), 55–64. Widener, M. J., & Horner, M. W. (2011). A hierarchical approach to modeling hurricane disaster relief goods distribution. Journal of Transport Geography, 19(4), 821–828. World Disasters Report. (2013). Focus on technology and the future of humanitarian action. Switzerland: IFRC. Yan, S., & Shih, Y. L. (2009). Optimal scheduling of emergency roadway repair and subsequent relief distribution. Computers & Operations Research, 36(6), 2049–2065. Yang, H., Yang, L., & Yang, S. H. (2011). Hybrid Zigbee RFID sensor network for humanitarian logistics centre management. Journal of Network and Computer Applications, 34(3), 938–948. Yi, W., & Kumar, A. (2007). Ant colony optimization for disaster relief operations. Transportation Research Part E: Logistics and Transportation Review, 43(6), 660–672. Yi, W., & Özdamar, L. (2007). A dynamic logistics coordination model for evacuation and support in disaster response activities. European Journal of Operational Research, 179(3), 1177–1193. Zhan, S. L., & Liu, N. (2011, April). A multi-objective stochastic programming model for emergency logistics based on goal programming (pp. 640–644), IEEE. Zhang, J. H., Li, J., & Liu, Z. P. (2012). Multiple-resource and multiple-depot emergency response problem considering secondary disasters. Expert Systems with Applications, 39(12), 11066–11071. Zografos, K. G., & Androutsopoulos, K. N. (2008). A decision support system for integrated hazardous materials routing and emergency response decisions. Transportation Research Part C: Emerging Technologies, 16(6), 684–703.

Customizable Tracking Solutions - Beacon Technologies

Models and Solutions for Radio Irregularity in Wireless ...

Customizable Tracking Solutions - Beacon Technologies, Inc.

virtual private networks technologies and solutions pdf

$pdf-149\armed-non-state-actors-in-international-humanitarian-and ...$

pdf-149\armed-non-state-actors-in-international-humanitarian-and ...

Chasing-Chaos-My-Decade-In-And-Out-Of-Humanitarian-Aid.pdf

Enabling Big Data Solutions with Centralized Data ... - Media15

Exact solutions for social and biological contagion models on mixed ...